Correlating enzymatic reactivity for different substrates using transferable data-driven collective variables.

Area of Science:

• Biochemistry
• Computational Biology
• Enzymology

Background:

• Identifying enzyme-substrate reactive conformations is crucial for understanding catalysis.
• Traditional methods face challenges due to complex free energy landscapes.
• Human pancreatic α-amylase is a key enzyme in type-II diabetes treatment.

Purpose of the Study:

• To apply machine learning (ML) techniques to identify enzyme reactive conformations (RC).
• To correlate ML-based collective variables (CVs) with experimental catalytic activity.
• To streamline computational modeling for enzymatic processes.

Main Methods:

• Utilized ML-based collective variables (CVs) to model enzyme-substrate interactions.
• Focused on human pancreatic α-amylase and malto-oligosaccharide substrates.
• Correlated the probability of being in an RC with experimental catalytic activity.

Main Results:

• Developed ML-based CVs that accurately predict reactive conformations.
• Demonstrated remarkable transferability of these CVs across different substrates.
• Significantly reduced computational demand and manual intervention in simulations.

Conclusions:

• ML provides an efficient approach to identify enzyme reactive conformations.
• This method advances the understanding of enzymatic catalysis.
• The approach holds potential for accelerating drug discovery and inhibitor development.